It is only procedures with 'SP_' prefix that cause the performance impact of scanning the system databases first.
'SP' should be fiine.
So, isn't some similar scanning occur when all the stored procedures start with the same prefix? ('SP' in this case). Not maybe for the system databases, but within the procedures in that specific context?
BOL only talks about not using 'sp_, another reason is that they may come up with the same name as your procedure on a future release.
Although I have not tested it properly, it seems that one should name his/her stored procedures as unique as possible for the maximum performance, starting from the first character.
I don't see how naming your stored procedures with the same prefix would affect performance at all. It has always been the norm in places I have worked to have a common procedure prefix as part of the naming standards.
Were any tests done on that?
I haven't done any tests my information comes from BOL.